URAP

Deep Learning and Statistical Reconstruction: Using virtual neural networks to image physical neural network architectures via magnetic resonance imaging

Roland Henry, Professor  
UC San Francisco  

Closed. This professor is continuing with Fall 2024 apprentices on this project; no new apprentices needed for Spring 2025.

UCSF’s Department of Radiology and Biomedical Imaging and Department of Neurology are excited to offer a combined educational and research opportunity for motivated undergraduate students in the medical imaging research team. 3D segmentation of structures in the brain and spinal cord is a problem that deep learning is uniquely equipped to solve. We are looking for aspiring data scientists and deep learning engineers to join our team to work on developing next-generation diagnostic techniques to monitor neurodegeneration in multiple sclerosis via deep learning on magnetic resonance imaging data.



We are working with one of the largest clinical imaging datasets for multiple sclerosis in the world with over 10 years of MRI data. Our laboratory specializes in quantitative analysis of images from MRI using image processing and machine learning. We work in close collaboration with the MS Clinic at UCSF Neurology to apply data science and deep learning techniques to uncover imaging biomarkers from clinical neuroimaging studies. We identify features that predict patient survival, therapeutic response, and other clinical outcomes.



We aim to build scalable annotation and segmentation workflows to intelligently compute standard radiological features such as spinal cord areas. We also want to build longitudinal models to analyze the evolution of neurodegeneration. Of particular excitement is building statistical model to quantitatively image the spatial pathology of MS. An undergraduate would help us by (1) building and training deep learning models on NVIDIA GPUs, (2) building scalable annotation platforms to further enrich the terabytes of existing data, (3) building infrastructure for managing the training and visualization of models for image segmentation.



Upon successful progress, it is expected that student submit/present at a national research meeting. Students are encouraged to seek out and apply for undergraduate research grants. Many of our students have went on to produce peer-reviewed publications.

Qualifications: Students from various majors are encouraged to apply, including but not limited to EECS, BioE, CS, data science, math, and statistics.

Day-to-day supervisor for this project: Amit Vijay Akula, Staff Researcher

Hours: 9-11 hrs

Off-Campus Research Site: * As this is a data-intensive project, much of the work can be done remotely at any place with a sufficiently fast internet connection * The day-to-day mentors for the project work at the Sandler Neuroscience Center at UCSF Mission Bay. * For the first semester, we do ask that a half-day (or sometimes a full-day) is spent in person at UCSF Mission Bay We spend the first semester on a "low hanging fruit project" and training. This semester will give undergraduates a broad exposure to projects in the center. As such, we usually ask undergraduates to consider a year-long commitment.

Return to Project List